Outline


why study synthesis?

computer-assisted learning: a history

quick fixes

ecology for a dynamic world

why study synthesis?


Scientific learning can be translated into tangible benefits
for society and the environment


Scientific learning can be translated into tangible benefits
for society and the environment


Data from: MJ Westgate, AIT Tulloch, PS Barton, JC Pierson & DB Lindenmayer (2017)
Optimal taxonomic groups for biodiversity assessment: a meta-analytic approach.
Ecography 40: 539-548


Scientific learning can be translated into tangible benefits
for society and the environment
…but this requires synthesis


L Bormann & R Mutz (2015) Growth rates of modern science:
A bibliometric analysis based on the number of publications and cited references.
JAIST 66(11): 2215-2222

computer-assisted learning:
a history

NR Haddaway & MJ Westgate (in prep)
A tool for predicting the time taken to conduct an environmental systematic review

  • The amount of published science is increasing rapidly
  • Current methods rely on intelligent people performing mundane tasks
  • We are rapidly approaching a ‘step change’, beyond which manual processing simply fails

R Borah, AW Brown, PL Capers & KA Kaiser (2017) Analysis of the time and workers needed to conduct systematic reviews of medical interventions using data from the PROSPERO registry. BMJ Open 7: e012545

quick fixes

stages of a systematic review

  • planning: setting the scope of the review, budget, methods etc.
  • searching: trawling academic databases and grey literature
  • screening: removing duplicates, identifying relevant material
  • analysis: working out what your papers tell you
  • reporting: writing up and publishing your results


    We can hasten (some of) these stages using existing methods

  • standardised import of multiple formats
  • flexible de-duplication
  • interactive visualisation and article selection

https://revtools.net

Total search results: 31,369 - Estimated number of unique articles: 18,433 - Number of pairwise links: 22,311

NR Haddaway & MJ Westgate (in prep)
A tool for predicting the time taken to conduct an environmental systematic review

other options

  • SR Toolbox: searchable repo for software options
  • robotreviewer: text mining, figure extraction
  • colandr: machine learning support for article screening
  • metagear: R package for (mostly manually) synthesis support

status of machine learning

  • already available: No need to (re)invent methods for ecology
  • efficient: testing shows a time reduction of up to 97%*
  • widely tested: topic models cited 20,859 times (published 2003)



*A O’Mara-Eves, J Thomas, J McNaught, M Miwa, S Ananiadou (2015)
Using text mining for study identification in systematic reviews: a systematic review of current approaches
Systematic Reviews 4(1): 5

ecology
for a dynamic world

100 years of ecology: what are our concepts and are they useful?

William R. Reiners, Jeffrey A. Lockwood, Derek S. Reiners & Steven D. Prager


Abstract. On the occasion of the Ecological Society of America’s centennial, we sought to learn which ecological concepts members value in terms of their utility. This required defining “concept” and selecting concepts from current ecology textbooks that might arguably belong to a normative set.

Ecological Monographs (2017) 87(2): 260-277

Progression of scientific terms

  • Novel: New term introduced
  • Pseudocognate: Appears obvious and intuitive
  • Ambiguous: Starts to be interpreted in different ways
  • Omnibus: Serves every use and user
  • Cluster concept: Requires careful disambiguation
  • Nonconcept: Disambiguation fails
  • Panchreston: Term means everything (and nothing)


    HEW Cottee-Jones & RJ Whittaker (2012)
    The keystone species concept: a critical appraisal.
    Frontiers of Biogeography 4(3): 117-127


MG Newberry et al. (2017) Nature 551: 223-226


The Australian, 24.08.2017


P Plaven-Sigray, GJ Matheson, BC Schiffler & WD Thompson (2017)
The readability of scientific texts is decreasing over time eLife 6: e27725


Vinkers CH, JK Tijdink & WM Otte (2015) BMJ 351: h6467

Benefits of word association networks

  • Network structure gives information on relative importance of terms, and can identify communities of related terms
  • Identifies ecological synonyms (econyms?); closely related terms for consideration in systematic reviews
  • Validation of machine learning algorithms
  • Can be used to inform discussion on the role of key concepts in ecology (e.g. species richness)

Summary

  • Computer-supported synthesis is expanding
  • It is already affecting ecological research
  • Enormous potential to harness these tools for creative research


~ special thanks to ~
David Lindenmayer
Don Driscoll
Neal Haddaway
Samantha Cheng

ARC Centre of Excellence
for Environmental Decisions

~ to contribute to ecoterms ~
online: https://martinwestgate.com
twitter: @westgatecology

~ this presentation was created using R ~
rendering: rmarkdown
analysis: revtools | tm | topicmodels | ade4
visualisation: plotly | ggridges | viridis